Re: [PATCH v1 02/19] xen/xsm: wrap around xsm_sysctl with CONFIG_SYSCTL

["Alejandro Vallejo" <alejandro.vallejo@cloud.com>]

On Thu Mar 13, 2025 at 12:05 PM GMT, Jan Beulich wrote:
> On 13.03.2025 12:43, Alejandro Vallejo wrote:
> > On Wed Mar 12, 2025 at 4:06 AM GMT, Penny Zheng wrote:
> >> --- a/xen/include/xsm/dummy.h
> >> +++ b/xen/include/xsm/dummy.h
> >> @@ -180,11 +180,18 @@ static XSM_INLINE int cf_check xsm_domctl(
> >>      }
> >>  }
> >>  
> >> +#ifdef CONFIG_SYSCTL
> >>  static XSM_INLINE int cf_check xsm_sysctl(XSM_DEFAULT_ARG int cmd)
> >>  {
> >>      XSM_ASSERT_ACTION(XSM_PRIV);
> >>      return xsm_default_action(action, current->domain, NULL);
> >>  }
> >> +#else
> >> +static XSM_INLINE int cf_check xsm_sysctl(XSM_DEFAULT_ARG int cmd)
> >> +{
> >> +    return -EOPNOTSUPP;
> >> +}
> >> +#endif
> > 
> > Doesn't this need to be -ENOSYS instead?
>
> There shouldn't be any ENOSYS outside of the top-level hypercall handlers.
> Granted we have many violations thereof, some of them not very reasonable
> to fix (for guests looking for the specific but wrong error code).
>
> Jan

That would be the case if the else branches were dropped. They have no use
AFAICS.

Cheers,
Alejandro